1. Field of the Invention
The invention relates to digital data transmission systems and to fiber optic digital data transmission systems.
2. Description of the Prior Art
Fiber optic digital transmission systems have been developed which have tremendous capacity to transmit large amounts of information over a single channel because of the large bandwidth which may be modulated upon an optical carrier signal. While the bandwidth of fiber optic transmission systems is a tremendous advantage in comparison to wire or radio frequency transmission mediums, digital fiber optic communication systems are nevertheless subject to problems caused by signal attenuation during transmission and noise.
The simplest form of detecting digital fiber optic communications is by using a threshold detector which senses every signal above a given threshold as a bit of information and every signal below the threshold as the absence of a bit. Threshold detectors are not able to reliably distinguish between high amplitude noise pulses and the transmitted digital data. Moreover, even though the optically encoded digital signal which is coupled to the fiber optic transmission medium from a transmitter or repeater may be relatively free from high amplitude noise, the fiber optic transmission medium may attenuate the digital signal sufficiently during transmission to drop the digital signal amplitude down close to or below the threshold level. For example, it is known that attenuations of the digital signal level in a fiber optic transmission medium of up to 20 dB or more may occur as a consequence of "poor[ quality splices which may be caused by a myriad of factors. To minimize digital signal attenuation in fiber optic transmission mediums, expensive connectors are used which require skilled labor to install. Moreover, expensive low noise threshold detectors are used to permit detection of attenuated digital signals. The difficulty in detecting attenuated digital signals transmitted by fiber optic transmission mediums requires the use of a larger number of repeaters to amplify the digital signal amplitude than otherwise would be required if threshold detectors could reliably detect low signal to noise ratio signals.
Current fiber optic communication systems use light emitting diodes which are operated at high power levels to produce a digital signal which has been modulated to a high amplitude. The operation of the light emitting diodes at high power levels can cause their premature failure. The operation of the light emitting diodes at high power levels is a consequence of the requirement that the amplitude level of the optical signal must be boosted to a sufficiently high level to permit accurate threshold detection and increased distance between repeaters.
In order to avoid the inherent unreliability of threshold detectors in accurately detecting the digital data which is being transmitted by a fiber optic transmission medium, sequenced voting devices have been used. In this approach, a digital signal is broken up into a number of slices, 16 for example, which are each "voted on" by single level or multilevel threshold detectors. If a certain number of threshold detectors affirm the presence of a digital signal, then a fixed digital signal is regenerated. The disadvantage of the sequenced voting devices has been their expense, and they are highly data-speed sensitive.
Satellite communications systems are extremely susceptible to problems caused by inaccurate detection of digital data at a transmitting station prior to transmission. Because of the approximate one-third of a second required to communicate between two ground stations via a relay from a geosynchronously orbiting satellite, any error in detecting a digital signal at a ground transmitting system, which is to be relayed to another ground station via satellite, after transmission will present the satellite communication system with a difficult error correction problem. To date, correction of detection errors which are discovered after transmission by a ground transmitting system have required the buffering of large amounts of data and sophisticated data processing because of the extremely high transmission bit rates which are characteristically used by current satellite multiplex communication systems. In the future when the number of ground stations, satellites and data rates are projected to dramatically increase for digital satellite communication systems, the requirement of accurately detecting digital data at ground stations prior to transmission will be even more acute because of the projected increase in information being transmitted. A data transmission system which economically detects and reduces the rate of transmission of erroneous data will lessen the amount of buffering and data processing equipment required to correct erroneous digital data transmission below that needed in current satellite systems. Since there is a tremendous advantage to communicating with ground stations via a fiber optic transmission medium because of bandwidth and cost considerations, a highly accurate detector for detecting fiber optically transmitted data at a ground station would be of great use in improving communication.
Currently, methods used to assist in error correcting are becoming progressively complex and expensive. One major network technique, time division multiplexing, allows separation of channels by an interval of time, but as data rates increase then not only are more accurate clocks necessary to determine accurate time intervals but accurate time synchronization between different points of a network can become overwhelmingly burdensome and failure prone. Another network technique that is tending to become burdensome in the existing prior art is the use of parity bits and/or address information bits that proceed or follow a stream/package of data bits. Parity bits in conjunction with protocol bits are now burdening data streams (especially when there may be hundreds of data initiating devices in a network) with overhead data that is mounting to 20% and even to 40% of the data being transmitted. All this overhead must be handled, rehandled and separated from the actual data.
Frequency shift keying is a known modulation technique for transmitting digital data which uses two discrete frequencies to encode the high and low levels of a digital signal. The signal format of frequency shift keying does not transmit a fixed amplitude component representative of the high level bit position and additional information such as the present invention. Systems using frequency shift keying are not compatible with existing digital data transmitting systems which detect PCM by threshold detection.